Amazon, Google, AI and us: Are we too close for comfort? Beyond facial recognition, we’re giving smart devices and platforms our intimate biometric details.
BY BRIAN BENNETT APRIL 4, 2019 5:00 AM PDT
Computers aren’t just getting smarter, they’re studying us more closely too. Whether it’s in the name of public safety, fraud protection or simple convenience, we’re feeding AI systems details that identify us and track our comings and goings.
Equipped with electronic ears, eyes and brains, a sea of devices now monitor human activity and characteristics. The list includes everything from smart speakers, laptops and phones, to even doorbells and locks. You may not have noticed this trend, but it’s real, and growing.
CNET has spent the last two weeks documenting the current state of facial recognition technology, but that’s just one method of gathering data directly from your person. Here are a few examples of how other biometric technologies have seeped into the fabric of daily life, and where things could go next.
Fingerprint readers everywhere
Devices that can read, capture and match fingerprints to specific individuals have been with us for decades. In 1969, the FBI began looking at ways computers could help facilitate fingerprint identification.
Back then the idea was to free as many FBI staff members as possible from this tedious, labor-intensive task. By 1975 the Bureau’s efforts, aided by the National Institute of Standards and Technology, led to a functional fingerprint scanner prototype based on capacitive sensors.
Now fast forward to 2013. That’s when biometrics truly became mainstream, designed for consumer convenience, not only serious crime fighting. That year Apple integrated its Touch ID fingerprint reader into the iPhone 6, 6 Plus, and 5S. It also placed the security system in its new iPad models.
Since then both Google and Microsoft have launched similar identification solutions. Examples include Pixel Imprint for Google’s Pixel phones, and Windows Hello available on Microsoft Surface laptops and tablets. Other device makers have adopted fingerprint reader technology as well. On this list of high-end handsets are the Motorola G7, OnePlus 6T, Samsung Galaxy S10, and Sony’s Xperia XZ3.
All these phones, save the OnePlus 6T and Galaxy S10, use capacitive readers. They harness the same principle that the FBI did to build its early fingerprint scanner. Capacitors in the scanner measure fingertips for slight differences in electrical conductivity. The scanner then forms a finger ridge pattern from these differences.
The OnePlus 6T is quite different. Its sensor is optical, essentially a tiny camera that takes photos of fingertips. Since it creates flat, 2D images, this reader is relatively easy to fool. Not so with the Galaxy S10’s ultrasonic sensor.
The S10 actively pings your finger with sound waves when it’s touched, These waves bounce off skin and back to the reader for analysis. With that data, it constructs a detailed 3D image that’s tougher to duplicate or spoof.
They know your voice
The smart speaker’s rise in popularity is undeniable. The number of US smart speaker users is expected to swell to 76.5 million by 2020. According to eMarketer, that’s up from 16 million in 2016, a compounded annual growth rate of 47.9%.
Amazon’s Echo products with Alexa onboard still represent the lion’s share of these gadgets. Google Assistant though is in hot pursuit. And as the companies compete for market share, the capabilities of both voice assistants has steadily increased.
These speakers don’t just listen out for your spoken commands. They can recognize, and distinguish between the voices of individual home members. That enables them to deliver personalized responses and results to match specific people.
It’s a neat trick that comes in handy when you want a rundown of your personal schedule, tasks, shopping lists and so on. For this to work, you’ll need to train these systems. And that means providing them with numerous voice samples. The audio snippets then find their way to remote servers for comparison, identification, fast retrieval and storage.
The eyes can’t lie
Like fingerprint readers and voice recognition, iris-scanning technology has slowly crept into ordinary life. The iris, the pigmented area of the eye outside the pupil, is unique from person to person. Even individuals with the same genetic makeup — identical twins, for instance — will exhibit distinct iris patterns dissimilar from each other. That makes the allure of this identity confirmation technique strong.
The potentially high speed and accuracy of iris scans is also compelling. A typical scanning system consists of a digital video camera, a low-energy infrared light emitter, plus hardware and software processing.
The scanner emits an invisible IR light beam that illuminates the iris briefly. The system records the iris colors and other unique features. It then breaks the pattern down into a 256 byte code. An unknown individual’s iris code is compared with a database of enrolled codes. The entire scanning, matching and recognizing process takes just a few seconds.
Some consumer devices have included iris-scanning features for years. Samsung chose Qualcomm’s 3D Sonic fingerprint reader for the Galaxy S10, but the Samsung Galaxy S7, S8 and S9 phones all shipped with iris-based identification systems.
In 2017, Qualcomm entered a license agreement with iris-based authentication company Eyelock. That could be a sign this access control method will soon find a home on Qualcomm mobile processors and platforms.
Two hearts don’t beat as one
Basic heart rate monitors have been a standard feature of fitness trackers and smartwatches for years. A few new devices such as the Apple Watch 4, take things a step further. Instead of merely recording heart rate, they also can take electrocardiogram (EKG or ECG) readings. This enhanced data allows the trackers to sense your heart rhythm.
EKG data is useful for many reasons. The first is greater awareness of heart health. Armed with these upgraded sensors, wearable devices can detect and alert you of potentially dangerous situations. That includes abnormal or irregular heart rhythm and even atrial fibrillation.
Another way heart rhythm is useful is its potential for biometric identification. According to company Bionym, each person’s heart rhythm and EKG characteristics are unique. Bionym says that’s a result of differences in heart size, position, shape, muscle conductivity and so forth. The Nymi band, a watch-sized wearable device, is Bionym’s heart rhythm authentication solution.
Bionym envisions the Nymi to serve as an ultra secure biometric key. Each Nymi is linked to only one individual user and their unique EKG identification profile. Worn by its owner, the Nymi is designed to offer authorized access for registered individuals only.
Potential applications include controlling entry to sensitive office and laboratory facilities, along with tools, equipment and digital accounts. At the moment the Nymi appears targeted at enterprise customers, though it’s conceivable products like the Apple Watch 4 might one day offer similar features to ordinary consumers.
Following your footsteps
It’s easy to see how physical features like fingerprints can identify individuals, but what about the way you walk? Researchers at Shinshu University in Tokida, Japan have examined exactly that.
They paid close attention to the way the foot makes contact with the ground while walking. Specifically they observed the timing of heel-strike, roll to forefoot and the final push off of the toes. They also recorded pressure levels exerted by the foot during the process.
Ultimately the scientists found that these dynamic foot pressure patterns were unique to specific individuals. In fact, they were able to identify particular people using the technique with an accuracy rate of 99.6%.
Wearable technology company Motiv plans to take gait recognition out of the realm of pure research. Its $199 Motiv Ring is packed with numerous sensors to measure physical activity and biometric data. It tracks the steps you take, heart rate, and sleep time. Motiv also touts the Ring as a frictionless security device, one that provides 24/7 authentication for online accounts and transactions.
Most interesting though, is the Ring’s WalkID feature. It uses the device’s internal accelerometer to figure out your individual gait, and determine whether you’re the person actually wearing the Ring. Currently the function remains in experimental beta. That said, it’s open to all Motiv Ring owners to give it a spin.